Sample size determination for BCI studies: How many subjects and trials?

Area of Science:

• Neuroscience
• Biostatistics
• Human-Computer Interaction

Background:

• Reliability of statistical results in brain-computer interface (BCI) studies is heavily influenced by sample size and statistical power.
• These factors are frequently overlooked during the planning and reporting phases of BCI research.
• Classical power calculations are often unsuitable for the nested experimental designs common in BCI research.

Purpose of the Study:

• To introduce simulation-based sample size determination (SSD) as a methodology for planning BCI studies.
• To demonstrate how SSD can determine the required number of subjects and trials for precise BCI accuracy estimation.
• To address the constraint of limited budgets in BCI research sampling.

Main Methods:

• The paper proposes and details the methodology of simulation-based sample size determination (SSD).
• This approach is designed to overcome the limitations of classical power calculations in nested BCI designs.
• The method is illustrated with examples relevant to BCI accuracy estimation under budgetary constraints.

Main Results:

• The proposed simulation-based sample size determination (SSD) method provides a viable approach for planning BCI studies.
• It enables researchers to calculate the necessary sample sizes (subjects and trials) for achieving a desired precision in BCI accuracy estimates.
• The method is effective even when study budgets impose constraints on the amount of data that can be collected.

Conclusions:

• Simulation-based sample size determination (SSD) is a flexible and powerful tool for BCI research planning.
• This methodology ensures reliable statistical results by adequately addressing sample size and statistical power.
• The generalizability of the method allows its application across various experimental designs and statistical frameworks in BCI research.